4 ms·
When I was young I would try to give confidence intervals for my estimates, and they would always be ignored.
by plasticchris 4y ago
When I was young I would try to give confidence intervals for my estimates, and they would always be ignored.
- skirmish 4y agoMe too, that's how I learned to take my honest time estimate and always multiply it by \pi.
- kuhewa 4y agoWow is that a thing? I've learned from a supervisor to always * by 3 (they had been doing it to my estimates for a while without telling me) but I reckon the extra is a good idea
- blamestross 4y agoSame. If only project management tools took distributions as inputs.
- Jtsummers 4y agoMany can, and they can also run simulations based on those distributions. Many PMs don't know how to use these tools, choose not to, or answer to managers who want the answer and not a fuzzy answer.
- chrisfreder 4y agoI started building a project management tool where each task is given a "type" to auto generate distributions based on past projects / tasks. Having each task just be some block of text makes it hard to estimate distributions for each task. I'm not sure if this is futile though. Nobody likes hearing about uncertainty even if it is true. Schedule delays are a company culture problem and not a tech problem. Used to piss off my manager by giving ranges and uncertainty for most estimates. Then picked something closer to the 80th percentile timeline only to be told it was always too long of an estimate. Had to switch to giving the best possible time things would finish by because delays were easier to explain than accurate deadlines.